SGI integrates eXludus Grid Optimizer in life sciences offering

Print Friendly, PDF & Email

From HPCwire

SGI logoeXludus Technologies… announced a reseller relationship with SGI, a leader in high-performance computing. SGI will embed eXludus Grid Optimizer software in its new BioCluster solution, developed specifically for the life sciences market.

…”The SGI BioCluster, based on the SGI Altix XE platform and including eXludus Grid Optimizer software, is an ideal solution for pharmaceutical, biotech and university researchers seeking a turn-key cluster that will accelerate their genomics and proteomics work, and easily integrate into their existing infrastructure,” said Deepak Thakkar, Ph.D., BioSciences Segment Manager, SGI. “Through testing conducted on a 64 processor configuration, it was observed that Grid Optimizer provided up to 90 percent performance gains on key life sciences applications, which allows SGI to deliver a system to our customers that will enable faster time to insight and discovery,” Dr. Thakkar added.

I think that the magic in the eXludus software is that it virtualizes and allows scheduling individual cores in the system which for certain kinds of workloads, particularly embarrassingly parallel ones like you’d find in genome searches, would lead to better performance.

But if you know different/better, leave a comment.


  1. They do a lot with datastaging technology, they can move data to a node faster than a straight scp or rsync. I believe they use some method of broadcasting the data across the network. For “very specific” jobs it can give a performance boost, I think blast type jobs they can make a difference. They have many other products as well.

  2. John is right in that our “magic” is a result of our virtualized resource allocation layer. But there are in fact two versions of our Grid Optimizer technology. One was specifically designed for embarrassingly parallel applications, as John points out, and one that also caters to any workload type where we can do just as well with 10’000 individual jobs submitted by 10’000 users.

    What Chris is referring to is our other product, RepliCator, that has been frequently used by Life Science centers. Although originally designed for re-provisioning large numbers of virtual images in parallel, it has also been used for pre-staging data and automatically synchronizing with any workload manager.

    What’s interesting is that both technologies can be combined so that very large problem / cluster sizes can be handled at peak performance.

  3. Benoit – thanks for clearing that up! You guys should do a 411 on eXludus for the site; check out past 411s to get a feel for the format by clicking on the categories menu under the header and selecting “The 411”.